dela3499 / numerical-integration-explorer Goto Github PK

When a user hovers over one of the error plots, the function evaluations associated with producing it should appear on the main plot, along with shading to represent the area that contributes to the integral approximation.

With the current design, users can hover over the error graphs to show the points that were evaluated to create the approximation. But they're all shown at once. I think that you get more insight when you show the algorithm working over time. This is easy enough to do since you've got the function evaluation handy, and ordered by their evaluation time.

This may require submitting data from each algorithm that's more complex than a list of function evaluations. It might need to take into account separate stages of the approximation. (This may only apply to the Romberg method, since all the other methods don't really build over time, except by adding additional function evaluates. )

When a user is dragging an equation parameter, a 3D plot should show the error vs. function evals for each allowable value of the parameter. For instance, if a user is dragging frequency up or down, there should be a 3D plot showing how error varies with frequency.

A user should be able to drag 2 handles for the main plot. One controls phase and DC offset, while the other controls amplitude and frequency.

Perhaps controls should be restricted to moving in only one dimension at a time. This would preserve the notion of varying a single parameter, which makes the 3D plots of error vs. function evals vs. parameter possible.

If a control is allowed to vary in two dimensions, then I'd require a 4D plot of error vs. function evals vs. param 1 vs. param 2.

Alternatively, I could eliminate the function evals dimension by setting it to a single value - perhaps set by the user, or just hardcoded.

Right now, when variables become negative, a new negative sign is added, which adds unnecessary space. If this is fixed, the characters will jump around less as the user changes variables.

Since EaselJS provides helper methods for objects drawn to canvas, it'll make several tasks easier. (normally, I'd only have access mouseenter and mouseleave events for the whole canvas, rather than individual objects drawn to it.)